One of the most outstanding events of the news last week was the case of Juan Luis de Soto, the man who lost his ID in 2019 and after two months some thieves used his document to buy three cars and ask for eight loans from the bank. It cost him God and help to prove that they were impersonating him. After the thieves, a whole mafia specialized in this type of crime was uncovered. As the frequent notices of the Spanish Data Protection Agency (AEPD) show, it is a type of scam that is not strange, neither in our country nor in other countries around the world.
This has led scholars such as the psychologist Guido corradi, to say that “Identification by photo is a measure of the last century”, a statement supported by a battery of peer-reviewed studies from recent years that corroborate that, despite the fact that we see it as a harmless process, given its ubiquity and the naturalness with which we apply it in our day-to-day lives, it is a important public and financial security hole: “We have experimental evidence that a) we are bad at recognizing photos and faces; b) we believe that we are not; c) the usual practice (customs, banks, etc.) worsens our recognition ”, says Corradi.
A study from the University of Louisiana shows that the comparison of faces for passports and IDs is highly fallible, with “Error rates of between 10 and 20% under ideal conditions of replicability”Which, they warn, does not happen in the real world when you have to do this type of facial review, when you can have huge queues of people and process 200 or 300 faces in a day.
In addition, according to their tests, if someone’s photo was tried to match another photo (for example, that of a DNI with the digital image of a security camera) repeatedly, respondents were more likely to be suspicious of the image already report the non-coincidence to a greater degree. But if the objective to be analyzed was infrequent and did not go through the system many times, the respondents obtained higher error rates, around 40%, even if the interviewers cross-asked them if they were sure of their choice.
A later work at the same University went a little further: the performance of three groups was analyzed: 800 notaries registered, 70 employees of the bank’s safe and 35 university students. The plan was to see what effects the veteran could have. Spoiler: none, or rather, none good. The highest correlation of success was not obtained by those with more experience, in reverse: the only relevant correlation that occurred in all the tests was that, the younger the subject surveyed, the higher the percentage of success, according to the researchers, perhaps because cognitive decline is a determining factor faced with this task.
This means that work experience in the field of facial recognition does not help to better recognize cases of identity theft. Louisianans acknowledge that previous work has shown a field where experience did work: forensic examiners apparently do better than novices, but this is because these workers do receive satisfactory training in the comparison of images and have better conditions to analyze the faces of the subjects, while in jobs such as notaries, security guards and customs, who work in rapidly changing environments, the only training given is that of life experience ( which, as we have seen, is of no help).
In fact, it could happen that this work practice is counterproductive: When there is a fraudulent identity theft due to an erroneous facial recognition, the employee never or almost never finds out that he has made that mistake (the scammer will not warn him of his mistake), and for that reason he never receives signals that This is a difficult or poorly performing task, which can lead you to become overconfident.
In almost all studies, scientists recognize that there is one type of person, the “Superidentifiers”, who are very good at evaluating faces and are so consistently throughout their lives. They are people who have that talent, but that talent is often not part of the skills evaluated to access those positions where they have to practice it so much.